How rare is a rare death?
Some of the numbers in the Ledger's Bizarre & Rare column are decently evidenced. Some are folklore with a citation bolted on. Here is which is which.
The Bizarre & Rare section is the most enjoyable part of this site and by a wide margin the least reliable. That is not an accident of sloppiness — it is a structural problem with the whole category, and it is worth explaining before going through the individual claims.
Why rare-death statistics are so bad
Common causes of death are counted through civil registration: a certificate, an ICD code, an aggregate. The system is imperfect but it is a system.
Rare causes fall through it. There is no ICD code for "struck by a falling coconut". A death like that is coded to blunt force trauma, and the coconut vanishes. So anyone wanting a global figure for coconut deaths cannot query a database — they must gather news reports, or extrapolate from a single regional study, or repeat a number someone else produced that way.
News aggregation has a specific and severe bias: it counts what gets reported, and reporting depends on novelty, language and media density. A shark attack in Australia is reported worldwide. An identical death in a country with less international press coverage may never appear in any dataset. So these figures undercount unevenly — and once published, they get repeated indefinitely without anyone revisiting the original.
Tier one: reasonably documented
Shark attacks (~10/year). Genuinely well tracked. The International Shark Attack File at the Florida Museum of Natural History has catalogued unprovoked attacks for decades, and annual fatalities are typically in the single digits to low teens. Sharks are unusually well counted precisely because they are frightening enough for someone to have bothered.
Lightning strikes (~4,000/year, and probably more). Lightning fatalities are well documented in countries with good weather services, and the global picture is genuinely uncertain. Published estimates range widely — some researchers have put the global toll as high as the tens of thousands, concentrated in agricultural regions where people work outdoors and shelter is scarce. The Ledger's 4,000 sits at the conservative end of a badly constrained range.
Elevator failures (~30/year) and roller coaster fatalities (~4/year). Both are essentially United States figures, tracked by workplace-safety and consumer-product regulators. They are decent numbers for the US and should not be read as global ones. A large share of elevator deaths, incidentally, are workers during maintenance rather than passengers.
Tier two: weakly sourced but plausible
Hippopotamus encounters (~500/year). The figure of around 500 deaths a year is repeated almost universally in "deadliest animals" lists, and hippos genuinely are dangerous — territorial, fast, and frequently encountered by people fishing or travelling by boat. What is missing is a primary source. The number appears to have entered circulation and stayed there. Treat it as an order-of-magnitude guess.
Fatal selfie attempts (~380/year). Selfie deaths have been studied, notably in a 2018 review that identified 259 deaths worldwide between 2011 and 2017 by trawling news reports — an average nearer 40 a year in that window, though later work using wider criteria has found more, and news-report methods certainly undercount. The Ledger's 380 is considerably higher than the published figures support.
Vending machine tip-overs (~2/year). This one traces to real medical literature: a 1995 study documented a series of deaths from vending machines toppling onto people rocking them. The figure is US-only, decades old, and machines have since been redesigned to resist it. Probably an overestimate today.
Falling icicles (~100/year). Usually cited in connection with Russian cities, where roof ice is a genuine seasonal hazard. No robust global figure exists.
Tier three: folklore
The three below have no sound evidence behind them. They remain on the site because they are part of the folklore of death and the site is partly about that folklore — but they should not be repeated as fact.
Falling coconuts (~150/year). The famous claim that coconuts kill 150 people a year — reliably introduced as "more than sharks!" — has no basis. It appears to have grown out of a 1984 paper in the Journal of Trauma describing coconut-related injuries treated at a hospital in Papua New Guinea over four years. That paper documented a small number of local injuries. It did not produce a global death estimate, and no such estimate exists. The "150 a year" figure seems to have been generated later for a shark-comparison press release and then repeated forever.
Right-hand gear accidents (~2,500/year). The claim that thousands of left-handed people die annually using right-handed equipment descends from research published around 1991 arguing that left-handers die substantially younger. The methodology was heavily criticised, and the leading explanation for the finding is a cohort artefact: for much of the twentieth century, left-handed children were pressured to write with their right hands, so older generations report as right-handed at much higher rates than younger ones. Compare ages at death between the groups and left-handers look short-lived — not because they die early, but because the old ones were reclassified. Later studies have not supported a meaningful lifespan gap.
Spontaneous human combustion (~1/year). Not a recognised cause of death. Forensic investigators generally explain the handful of celebrated cases through the wick effect: clothing acts like a candle wick and body fat like the wax, letting a small external ignition source — a cigarette, a heater — sustain a slow, intensely localised fire for hours. It burns the body thoroughly while scorching remarkably little of the surrounding room, which is exactly the detail that made the phenomenon seem supernatural. There is no evidence of anyone igniting without an external source.
Why we get this so wrong
The reason these figures circulate is not that people are gullible. It is that human risk intuition runs on availability — we judge how likely something is by how easily an example comes to mind, and vivid, novel, heavily reported dangers come to mind very easily indeed.
Sharks kill perhaps ten people a year. Cardiovascular disease kills roughly eighteen million. The ratio is close to two million to one, and almost nobody's felt sense of danger reflects that. Meanwhile the genuinely large risks — a poor diet, a sedentary decade, an untreated blood pressure reading, a cigarette habit — arrive without any narrative at all.
Which is, in the end, the joke the Bizarre & Rare section is making. You are watching a coconut counter. The cardiovascular counter above it has moved several hundred times in the same period, and you almost certainly did not look at it.
Corrections to any of the above are genuinely welcome — particularly if you have a primary source for a figure listed here as unsourced. Send it over and this page will be updated.